How far away are we from that point, I think it is borderline happening already but with the help of human intervention proof reading. If a super-spammer wants to do it, eventually they will be noticed but if it is done more carefully and not automatically posted but a human proof reads the work to add a human touch to it, then becomes a little less detectable.
There are already huge accounts farmers operating in the forum, they have been for many years. They are beginning to change behaviours to adapt to the modern system (bots, AI, etc) and when one changes their style, others will follow. What will happen to the account farmers when one account gets caught? Nothing. They will continue to post with the other accounts until they get caught and in the interim will continue to create new accounts.
I wonder what percentage of active accounts in the previous 3 months are using AI and if it were possible to know, to use that as a comparator to the same period one year ago. That would be an interesting statistic.
What I fear about this is that the day will come when you will be able to tell an AI: "write a text about X but that it cannot be detected as written by an AI" and it will do it correctly, in an undetectable way. Then account farmers like this will be able to make gold, although I don't know what effect it will have on everything globally. For example, if when visiting a forum you can't be sure whether you are talking to humans or to text massively written by AIs, I don't know if people will lose interest in the forums.